OEM SCSI Disk Drive Type DGHS ULTRASTAR 18XP/9LP Hardware/Functional Specification 18.2GB and 9.1GB Models., 7200 RPM Version 4.50 Document Number AS06-0140-00 March 20, 1998 IBM Storage Systems Division 5600 Cottle Road San Jose, CA. 95193 Source filename=GBSZ7_14 Page 1 of 90
90
Embed
OEM SCSI Disk Drive Type DGHS ULTRASTAR 18XP/9LP 18.2GB ...ps-2.kev009.com/ohlandl/IBM_HD/dghs_spfw.pdf · DOCUMENT SUBJECT TO CHANGE WITHOUTNOTICE ULTRASTAR 18XP/9LP Hardware/Functional
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
OEMSCSI Disk Drive Type DGHS
ULTRASTAR 18XP/9LP Hardware/Functional Specification18.2GB and 9.1GB Models., 7200 RPM
This documentcontains theHardware/FunctionalSpecifications for theULTRASTAR 18XP/9LPHigh Performance Family of3.5-inchDisk Drives.
The productdescription andother data found inthis documentrepresentIBM's design objectives and isprovided for information and comparative purposes. Actualresults may vary based on avariety of factorsand the information herein is subject tochange. THISPRODUCT DATA DOES NOT CONSTITUTE AWARRANTY, EXPRESS OR IMPLIED. QuestionsregardingIBM's warranty terms or the methodologyused to derive the data should be referred to your IBMrepresentative.
Trademarks
IBM, NoID, and PredictiveFailure Analysis aretrademarks of International Business MachinesCorporationin the UnitedStatesand/or othercountries.
| − Low Voltage DifferentialSCA-2/80 pin and 68 pinRotary voice coil motor actuatorClosed-loopdigital actuatorservoEmbeddedsector servoMagnetoresistive (MR) heads16/17 rate encodingPartial ResponseMaximum Likelihood (PRML) datachannel with analogfilterNoID sectorformatAll mounting orientations supported1MB segmented cache bufferJumperabledrive suppliedterminator power (on some models)Jumperable on boardactiveSCSI terminators (optional on some models)
Interface Controller FeaturesMultiple initiator supportIntegrateddifferential tranceivers (on differential models only)Variable logical block lengths (512 - 732 supported)Nearly ContiguousReadRead-aheadcachingAdaptive caching algorithmsWrite CachingBack-to-backwrites (merged writes)Tagged and untaggedcommandqueuingCommandreordering (4 user selectablealgorithms)Automatic retry and data correction on read errorsAutomatic sector reallocationIn-line alternate sector assignmentDown-loadablefirmwareCustomizing controller jumpers
For example:-− Auto spindlemotor start− Auto Start Delay− Disable Target InitiatedSynchronous Negotiation− Disable UnitAttention− DisableSCSI Parity− Write protection
Reliability, Availability and Serviceability FeaturesSelf-diagnostics onpower upDedicated headlandingzoneMagnetic actuatorlatchEntire Read/Write customer data path protected by 32 Bit CRC18 Byte Error CorrectingCode(ECC)9 Byte ECC on the flyPredictiveFailure Analysis (PFA)Error Recovery Procedures(ERP)Data Recovery Procedures(DRP)Probability of notrecoveringdata:10 in 1015 bits readNo preventativemaintenance requiredEvent logging and analysisHigh temperature monitoring andlogging
The ULTRASTAR 18XP/9LP disk drive is available invarious models with avariety of the followingoptions.
| Capacity of 9.1GB or 18.2GB
Pleaserefer to section2.3, “Capacities by FormatLength” on page 14 for exact capacities based on userblock size.
Physicalconnectorsavailable
− 50 pin connectorsoffer an 8 bit bususing theSCSI 'A' connector− 68 pin connectorsoffer an 8/16 bit(Narrow/Wide) bususing theSCSI 'P' connector− 80 pin connectorsoffer an 8/16 bitSCSI bususing the'SCA-2' connector
Electrical Interfaces available
− SingleEnded(50, 68, and 80pin)− Differential (68 pinonly)
| − Low Voltage Differential(LVD) (68 and 80 pinonly)
The varioussupported combinations of the above options and the ModelNumbers thatcorrespond to thoseavailablecombinations arelisted in the followingtable.
Several of themodels listed in theabove table haveoptions that are notidentified via themodel number.They are asfollows...
Bezel (optional on some models)Jumperableactive terminators (on 68 & 50 pinsingle-ended models only).
All specificationnumbers are mean populationvalues unless otherwisenoted.
2.1 GeneralNote: The recordingband located nearest the diskouter diameter (OD) is referred to as'Notch #1', therecordingband located nearest the inner diameter(ID) is called 'Notch #16'. 'Average'values are weightedwith respect to thenumber of LBAs per notch when thedrive is formatted with 512 byte blocks.
Data transfer ratesNotch #1 Notch #16 Average
Buffer to/from media 22.40 11.52 18.27 MB/s (instantaneous)
Buffer to/from media 15.47 7.75 12.66 MB/s (sustained)
Host to/from buffer up to 40.0 MB/s(synchronous Ultra)up to 80.0 MB/s(synchronous Ultra2)
Rotational speed(RPM) 7200
Average latency (milliseconds) 4.17
Track Density (TPI) 8356
Minimum Maximum
Recording density (bpi) 132,670 150,000
Areal density (Megabits/square inch) 1109.0 1253.4
1.0" 1.6"
| Models 9.1GB 18.2GB
| Disks 5 10
| Heads 10 20
SeekTiming in milli seconds (measured at nominalvoltage andtemperature)1.0" 1.6"
The following voltage specificationsapply at thedrive power connector. There is nospecialpower on/off| sequencing required. ForDifferential SCSI modelssee: 2.4.8, “Additional 5V Current Requirements for| High Voltage Differential SCSI” on page 24 and 2.4.9,“Additional 5V Current Requirements for Low| Voltage Differential(Ultra 2)” on page 25.
Input Voltage+ 5 Volts Supply 5V (± 5% during run andspin-up)+ 1 2 Volts Supply 12V (± 5% during run) ( + 5 % / -7% duringspin-up)
Power Supply On/Off Requirements
+ 5 V 4.5 V/secMinimum Slew.+12V 7.4 V/secMinimum Slew.
There are no power offslew rate requirements.
Population PopulationMean Range
Power Supply Current 9.1GB Model
+5VDC (Power Save Mode) 0.63 Amps1 + / - 10 %
+5VDC (idle avg) 0.78 Amps2 + / - 10%
+5VDC (R/W baseline) Typical 0.85Amps + / - 10%
+5VDC (R/W pulse) Peak 1.25 Amps + / - 10%
+12VDC (Idle) 0.45 Amps + / - 10%
+12VDC (seek peak) Peak 2.07 Amps Maximum
+12VDC (Start Current) Maximum 1.6 Amps3 + / - 0.3
1 Power Savemode isautomatically invoked after 1 second ofinactivity, except whenread ahead isactive. In whichcasePowerSaveMode is invoked after 40 seconds ofinactivity.
2 5 Volt Current isgiven with termination power provided by theusing system.
3 The StartCurrent is thetotal 12 volt current required by the Drive.
Table 5. Power Measurementsmadeusing Clarke Hess Model 259Digital Wattmeter
Note:
For the purpose ofcalculations0.061 Watts per op can beused for the9.1GB and0.064 Watts per op for18.2GB This is not completely accurate because of the non linearscale but can beused forapproximations.
For thesemeasurements an op isdefined as a read transfer of 4k Bytes.Measurementswere taken about 5minutesafter file was spun up toallow the 12 voltpower readings to stabilize.
Example
If 9.1GB power wasrequired for 90 ops per second the calculationwould be asfollows
Idle power = 9.20 Watts
90 ops/second = 90 *0.061 = 5.49Watts
Total Power = 9.20 + 5.49 =14.69Watts
2.4.2.1 Differential ModelsThe powerincrease for differentialmodels will bebetween 0.5 watts atidle to 1.0watts for 60 ops/s.
| 2.4.2.2 Low Voltage Differential Models| The powerincrease for low voltage differentialmodels will bebetween 0.1 watts atidle to 0.2 watts for 60| ops/s.
2.4.3 Power Supply Current Profiles
These graphswere obtained bymeasuring apopulation of drives from the manufacturingprocess. Theresults are the averagefrom this population taken at nominalvoltageconditions.
All power supplies arenominal.
The results exclude inductive spikes caused by leads,power supplies andcomponents thatwill vary withdifferent setupconfigurations.
The graphsshown on thefollowing pages areapproximate representations of the power supply currentsbased on lab measurements.
2.4.4 Power Supply RippleExternally Generated Rippleas seen atdrive power connector
Maximum Notes
+5VDC 200 mV 0-20 MHzpeak-to-peak
+12VDC 200 mV 0-20 MHzpeak-to-peak
During drive start up andseeking, 12volt ripple is generated by thedrive (referred to asdynamic loading). Ifseveral driveshave their power daisy chained together then the power supplyripple plus other drive'sdynamic loadingmust remain within the regulation tolerance window of + / - 5%. Acommon supply withseparatepower leads toeachdrive is amore desirablemethod ofpower distribution.
2.4.5 Input CapacitanceInternal bulk capacitanceas seen atdrive power connector
+5VDC 72 ± 20% microfarad
+12VDC 510 ± 20% microfarad
2.4.6 Grounding Requirements of the Disk Enclosure
The disk enclosure is atPower Supply groundpotential.
From anElectroMagnetic Compatibility(EMC) standpoint itwill, in most cases bepreferable to provide acommon ground connection between thedisk enclosure and thesystem'smounting frame. With this inmind, it is important that thedisk enclosure notbecome anexcessivereturn current path from thesystemframe to power supply. Thesystem framemust bewithin + / - 150 millivolts of the drive's power supplyground. At no time should more than 35milliamps of current (0 to 100Mhz) beinjected into the diskenclosure.
2.4.7 'Hot Plug/Unplug' support
The term 'Hot Plug', refers to theaction of mechanicallyengaging a device topower and / or SCSI buswhen otherdevices may be active on thesame bus.
A comprehensiveclassification of thestate of theSCSI busduring this event is located in the SCSI-3 ParallelInterfaceStandard, Annex'A'.
Note: Case 3 isdefined as'Current I/Oprocesses not allowed during insertion orremoval'
Case 4 isdefined as'Current I/Oprocesses allowed during insertion orremoval'
While every effort was made to design the drive not to influence theSCSI busduring these events, it is asystem responsibility to insure voltage regulation andconformance to operational and non-operational shocklimits.
During Hot Plug events thenon-operational shocklevels should not beexceeded. Theoperational shocklevels of adjacentdrives should not beexceeded as well. Therecommended procedure is to prohibitwriteoperations to adjacentdrivesduring the HOTPLUG andduring the HOTUN-PLUG actions.
Reference section7.2, “Vibration and Shock” onpage 79 for theOperating andNon-Operating Shocklimits.
During Hot un-Plug the operational shocklimit specifications should not be exceeded. Ifthis cannot beguaranteed then thedrive should beissued aSCSI Stop Unitcommand that isallowed to completebeforeun-plugging. The basicrequirement isthat while the drive isoperational or spinning down (as aresult of aUNIT STOP orUn-Plug) the operationalshock limits are in effect. Once thedrive hascompletelystopped(15 secondsmax) the non-operational shocklimits are in effect. Therecommended procedure is toallow theun-pluggeddrive to rest in the drive bay for aminimum of 15seconds and then complete the removal.
During Hot Plug or Unplugevents thepower supplyripple on adjacentoperationaldrives should not gooutside of the+ / - 5 % regulation tolerance.
2.4.7.1 (SCA-2 models)
Based on theconnectorclassification called out inSFF-8046, theULTRASTAR 18XP/9LP 80pinSCA-2 drive is 'Case 4' compliant, when thesystem has properlyimplemented the SFF-8046guidelines.
2.4.7.2 (50,68 pin models)
Based on theSCSI Parallel Interface classification, it isrecommended that theusing systemcomply with'Case 3'guidelines toeliminate the chance ofaffecting an activebus.
In systemsthat cannotafford to quiesce theSCSI bus, but can meet the requirements ofvoltage regulation,operational and non-operational shock, thefollowing guidelines arerecommended tominimize the chance ofinterfering with theSCSI Bus:
Plug
1. Common ground should be made betweendevice andpower supply ground
2. Plug device onto the bus
3. Power updevice ( no special sequencing of 5 or 12 volts)
4. Device is ready to beaccessed
Un-Plug
1. Power downdevice (no special sequencing of 5 or 12 volts)
2. Un-plugdevicefrom the bus
3. Remove commonground
2.4.8 Additional 5V Current Requirements for High Voltage Differential SCSI
PopulationAdditional Power Supply Current Notes Mean
A full Bring-Up Sequenceconsists of aPower-upSequence and Start-Up Sequence, as Figure 7 shows.
"Power On" isdefined as when thepower at thedrive meets all of the powerspecifications as defined inthisdocument.
The Start-Up sequence spins up the spindlemotor, initializes the servo subsystem, performsBasic-Assurance-Tests-2(BATs2) (verifies read/write hardware), resumes "Reassign inProgress"operationsand more. See theULTRASTAR18XP/9LP SCSI Logical Interface Specificationfor additional details onthe Start-Upsequence.
If a SCSI Reset isissued while the drive is in either aPower-Up or Start-UpSequence,that same Sequencestarts again. In allother cases when aSCSI Reset isissued thepresent state of themotor is notaltered.
Reference3.10.1.3,“Start/Stop Unit Time” onpage 43 for additionaldetails.
A startup sequence initiated bySCSI "Start/Stop Unit"command thatfollows a spindlestop initiated by aSCSI "Start/Stop Unit"command byless than 10seconds, may result in theStartupsequenceincreasing byas much as 10 seconds. For example, if adelay of only 3secondsexistsbetween the 2commands, the 2ndcommand can take 7seconds longer than if 10 seconds ormore had beenallowed between the"Start/StopUnit" commands.
2.5.1 Spin Down TimesAfter power is removed thedrive should beallowed 15 seconds topark theheads and spin down before anyattempt is made tohandle thedrive.
It is recommended thatafter power is removed, a period of 2seconds should be allowed beforepower isreapplied to thedrive.
In the event of apower glitch the drivewill normally execute aPower On Reset and then go throughit'sPower-Upsequence.Depending on the duration of theglitch the drive mayspin down and then spin backup or may resetitself in which case thespindlemotor is turnedoff. If the host system detects apower glitchit is recommended that a SCSI BusReset be performed. Thisallows the drive to bebrought to a Readycondition in a controlledmannerfrom a known state.
Drive performancecharacteristics listed inthis chapter aretypical valuesprovided for information only, sothat theperformance for environments and workloadsother than those shown asexamples can beapproxi-mated. Actualminimum and maximumvalues will vary dependingupon factors such asworkload,logicaland physicaloperating environments and manufacturingprocess variations.
3.1 Environment Definition
Drive performance criteria is based on thefollowing operating environments. Deviations fromthese environ-ments maycause deviationsfrom values listed inthis specification.
Nominal physicalenvironment(voltage,temperature, vibration,etc.) as defined elsewhere inthis specifi-cation.
Block lengths areformatted at 512 bytes per block.
The number of databuffer segments is 16. Thetotal databuffer length is671KB. The size of eachequally sized segment, in either bytes or blocks, is determined via theSCSI Mode Page 8h parametercalled "CacheSegmentSize".
Ten byte SCSIRead andWrite commands areused.
SCSI environmentconsists of asingle initiator andsingle target with noSCSI Bus contention.
Buffer full/empty ratios are set to theiroptimum valuessuch that a minimum number ofintermediatedisconnects occur during theSCSI datatransfer and the overlap of theSCSI anddisk data transfer ismaximized. This minimizesCommandExecution Times with no bus contention.
The initiator delay while transferringSCSI command,status,message, and data bytes isassumed to bezero.
TaggedCommandQueuing is notused, unless otherwisespecified.
All Current Mode Parameters are set to their Defaultvalues except wherenoted.
SCSI datatransfers are successfullynegotiated to be 40MB/s.
Averages arebased on a samplesize of10,000operations.
3.2 Workload Definition
The drive'sperformance criteria is based on thefollowing commandworkloads. Deviations fromthesework-loads may cause deviationsfrom this specification.
Operations are either all Reads or allWrites. The specifications forCommandExecution Time withRead Ahead describeexceptions to this restriction. Forthat scenario allcommands arepreceded by aRead command,except for sequential writecommands.
The time between the end of an operation, and when the next operation isissued is 50 ms, + / - arandomvalue of 0 to 50 ms, unless otherwisenoted.
3.2.1 SequentialNo Seeks. The target LBA for all operations is the previous LBA + Transfer Length.
3.2.2 RandomAll operations are torandomLBAs. The average seek is an average weighted seek.
3.3 Command Execution TimeCommand Execution, orService,Times are the sum ofseveral BasicComponents. Those Components are -
1. Seek2. Latency3. CommandExecution Overhead4. DataTransferto/from Disk5. DataTransferto/from SCSI Bus
The impact or contribution of thoseBasic Components to CommandExecution Time is afunction of theworkload being sent to thedrive and theenvironment in which thedrive is beingoperated.
The following graphsshow CommandExecution Times for fourgenericworkloads
SequentialReadsSequential Writes
RandomReadsRandomWrites
with several differentrequested Transfer Lengthswhile running in various environments whose keyfactorsare identifiedundereach graph.
| Note: Times arecalculated with TypicalData Sector TransferRates for18.2GB models in Ultra mode and| are averaged over the entiredrive.
Note: In the following Graphs,"TCQing" meansTaggedCommandQueueing.
The average timefrom the initiation of theseek, to theacknowledgementthat the R/Whead ison the trackthat contains thefirst requestedLBA. Values arepopulationaverages, and vary as afunction of operating conditions. Thevalues used in thegraphs showingCommandExecutionTimes for sequentialcommands is 0 ms and thevalues for random commands are shown insection2.1, “General” onpage 12.
LatencyThe average timerequired from the activation of the read/write hardware until the targetsectorhas rotated to the head and theread/write begins. This time is 1/2 of arevolution of thedisk, or4.17 ms.
Command Execution OverheadThe average timeadded to theCommandExecution Time due to theprocessing of theSCSIcommand. It includes all time the drivespends processing acommandwhile not doing a diskoperation or SCSI datatransfer, whether or not it is connected to the bus.(See 3.6,“ReadCommandPerformance” onpage 38 and 3.7, “Write CommandPerformance” onpage 39 forexamples of detailed descriptions of thecomponents ofCommandExecution Overhead.) Thevalue of thisparametervaries greatlydependingupon workloads and environments.
The following values areused when calculating theCommand Execution Times.'RA' meansReadAhead isenabled.
Table 7. OverheadValues. (All times are in milliseconds.)
Other Initiator controlled factors such as use of disconnects,TaggedCommand Queueing andthe setting ofMode Parameterslike DIMM, DPSDP andASDPE also affect CommandExe-cution Overhead.They also affect SCSI Bus Overhead which ispartially a subset ofCommandExecution Overhead.
SCSI BusOverheadis defined as the time the device isconnected to the bustransferring allSCSICommand, Status andMessagephase information bytes. Thisincludes any processing delaysbetween SCSI Busphaseswhile remaining connected to theSCSI Bus. Initiatordelays whiletransferringinformation bytes are assumed to be zero. This time does not include theSCSI Dataphasetransfer. (See 3.6,“Read CommandPerformance” onpage 38 and 3.7, “Write CommandPerformance” onpage 39 formore detailed descriptions of thecomponents of SCSI BusOver-head.)
Post Command Processingtime of 0.1 ms is defined as the average timerequired for processcleanupafter thecommand hascompleted. If a re-instruct periodfaster thanthis time is used, thedifference isadded to theCommandExecution Overhead of the next operation.
Data Transfer to/from DiskThe average timeused to transfer the data between the media and the drive's internal databuffer.This is calculated from:
There are four interpretations of MediaTransferRate. How it is to beused helpsdecidewhichinterpretation is appropriate touse.
1. InstantaneousData TransferRate
The same for agiven notch formatted at any of the supportedlogical block lengths. It variesby notch only anddoes not include any overhead. It is calculated from:
1/(individual byte time)
2. TrackData Sector TransferRate
Varies dependingupon theformattedlogical block length andvariesfrom notch to notch. Itincludes theoverhead associated with each individual sector. This is calculated from:
(user bytes/sector)/(individual sector time)
(Contact an IBM CustomerRepresentative forindividual sector times of the various for-matted blocklengths.)
Note: Theserates are used to help estimateoptimum SCSI Buffer Full/Empty Ratios.
3. TheoreticalData Sector TransferRate
Also includes time required fortrack andcylinder skew andoverhead associated with eachtrack. Use thefollowing to calculate it.
Data SectorTransfer Rate =
Bytes/cylinder
time for 1 cyl + trackskews + 1 cyl skew
4. Typical Data Sector TransferRates
Also includes theeffects of defectivesectors and skipped revolutions due to errorrecovery.(See Appendix B. of theULTRASTAR18XP/9LPSCSILogical Interface Specificationfor adescription of errorrecoveryprocedures.)
Rates fordrivesformatted at 512 bytes/block are located inTable 8 onpage 34.
Note: The values for TypicalData SectorTransfer Rates assume atypically worst casevalue of 3 errors in 109 bits read atnominal conditions forsoft error rate.
Note: Contact an IBM CustomerRepresentative forvalues whenformatted at otherblock lengths.
Note: Each group ofcylinders with adifferent number ofgrosssectors per track iscalled anotch. "Average" valuesused in thisspecification aresums of theindividualnotch values weighted by thenumber ofLBAs in the associatednotches.
Data Transfer to/from SCSI BusThe time required to transfer data between theSCSI bus and thedrive's internal databuffer, thatis not overlapped with the time for theSeek,Latency orData Transferto/from Disk. This timeis based on aSCSI synchronous datatransfer rate of 40.0MB/s.
The SCSI datatransfer rate isdependent on the mode, either synchronous or asynchronous. Italso dependsupon thewidth of the data path used. 8 and 16 bittransfers aresupported.
When thedrive is configured for an 8 bit wide transfer asynchronous datatransfer rate of 20MB/s can berealized. The 16 bit widemaximum synchronous datatransfer rate is 40MB/s.
| For Low Voltage Differential drivessynchronous datatransfer rates of 20MB/s for 8 bit wide| transfers, and 80MB/s for 16 bit wide transfers can be realized.
The asynchronous datatransfer rate isdependent on both the initiator and targetdelays to theassertion and negation of theSCSI REQ and ACK signals. It is alsodependent on SCSIcabledelays. The drive iscapable of supporting up to 5 MegaTransfer/second(MT/s) asynchronousdata transfer rates.
The SCSI datatransfer rate specification only applies to theData phase forlogical block data forRead, Write, Write and Verify, etc... commands. The datarate for parameter/sense data forRequestSense,Mode Select, etc.commands is notspecified.
3.3.2 CommentsOverlap has been removed from theCommand Execution Time calculations. Thecomponents of theCommandExecution Times are trulyadditive times to the entireoperation. Forexample,
The SCSI Bus Overhead data is notincluded in the calculationsince some of its components arealsocomponents ofCommand Execution Overhead and the remainingcomponentsoverlap the DataTransferto/from Disk. (See 3.6,“Read CommandPerformance” onpage 38 and 3.7, “Write CommandPerformance” onpage 39 for details.)
The PostCommandProcessing times are notcomponents of theCommandExecution timethereforethey are not included in the calculation ofenvironmentswhere the re-instruct periodexceeds thePostCommandProcessing time.
With Read Ahead enabled, this specification measures aRead orWrite command when theimmediatelyprecedingcommand is a Read command(which starts up theRead Ahead function). If theprecedingcommand is aWrite command, then the timedifference due toReadAhead iszero.
Longer inter-opdelay, or low re-instruction rate,environments are suchthat the ReadAhead function hasfilled the drive's internal data cache segment before the nextRead orWrite command isreceived.
Environments with inter-opdelaysless than 1revolution period, or high re-instructionrates, aresuch thatthe ReadAhead function isstill in the process offilling the drive's internal data cache segment when the nextRead orWrite command isreceived. Forsequential reads,CommandExecution Time is 1revolution lessthan similaroperations with equal inter-opdelays andReadAheaddisabled.
The effects of idle timefunctions are not included in theaboveexamples. The sections3.2.1,“Sequential” onpage 29 and 3.2.2,“Random” on page 29 both defineenvironmentswhere the effects due toincreasedcommandoverhead of IdleTime Functionsupon CommandExecution time areless than0.07%.
3.4 Disconnection During Read/Write Data Phase
If a nonzeroMaximum Burst Size parameter isspecified, the drivedisconnectsafter transferring thenumberof blocks specified by theMaximum Burst Size parameter. This disconnectionrequiresapproximately 6 µsand the subsequent reconnectionrequiresapproximately 15 µs.
The drive alsodisconnects prior tocompletion of theData phase if thedrive's internal databuffer cachesegment becomesempty during a Read command orfull during a Write command. Thisdisconnectionoccurs regardless of theMaximum Burst Size parameter. This disconnectionrequiresapproximately 6 µsand the subsequent reconnectionrequiresapproximately 15 µs.
3.5 Approximating Performance for Different EnvironmentsThe values for several BasicComponents maychange based on the type ofenvironment and workload. Forexample,CommandOverhead may change because certain internalcontrol functions may nolonger be over-lapped with either the SCSI ordisk data transfers, etc. The followingparagraphs describewhich parametersare affected bywhich features.
For read commands with ReadCaching EnabledCommand Execution time can be approximated bydeleting Seek,Latency andData Transfer to/from Disk times from those shown on thegraphs if all of therequested data isavailable in a cachesegment (cache hit).When some, but not all, of the requested data isavailable in a cachesegment (partial cache hit)Data Transfer to/from Disk will be reduced but notelimi-nated.Seek andLatency may or may not be reduced dependingupon thelocation of requested data not inthe cache and location of the read/write heads at the time thecommand wasreceived. Thecontribution ofthe DataTransferto/from SCSI Bus to theCommandExecution time mayincrease since a larger, orentire,portion of thetransfer may no longer be overlapped with thecomponents thatwere reduced.
3.5.2 When Read Ahead is Enabled
The reduction in sequential (contigous andnon-contiguous)read workload with long inter-opdelaysCommandexecution times can beapproximated byusing the followingequation:
The magnitude of the performance advantage of theReadAhead with opdelays of 0 to 1 rev varies with thesize of thedelay. Since the range of delays isless than the time for onerevolution, the operation is"synchro-nized to the disk". TheReadAheadsavings can beroughly approximated by:
DELAY - (time for one revolution) = ReadAheadsavings
Note: This time alsovaries with thesize of the datatransfer due to thedifferencebetween the SCSI datatransfer rate andDisk datatransfer rate. This time is insignificant for a0.5KByte transfersize and hasbeenignored in theabove equation.
3.5.3 When Write Caching is Enabled
For write commands with theWrite CachingEnabled(WCE) Mode parameter bitset,CommandExecutiontime can be approximated bydeleting Seek,Latency andData Transfer to/from Disk times from thoseshown in thegraphs. Thecontribution of theData Transferto/from SCSI Bus to theCommandExecutiontime may increase since a larger, orentire, portion of the transfer may no longer be overlapped with thecomponents thatwere reduced.
The reduced timeseffectively areadded to thePostCommandProcessingTime.
Like TaggedCommandQueuing, the potential toreduceCommandExecution Overheadexists due tocon-currentcommandprocessing.
Like TaggedCommandQueuing, when the WCE bit is set Back-To-Backwrite commands are supported.See 3.5.5.2,“Back-To-BackWrite Commands” onpage 37 formore information.
3.5.4 When Adaptive Caching is Enabled
The Adaptive Caching featureattempts toincreaseRead Cache hit ratios bymonitoring workload andadjusting cachecontrol parameters, normally determined by theusing system via theSCSI Mode Parame-ters, with algorithms using thecollectedworkload information.
The effects ofCommandExecution Overhead can bereducedsignificantly if TaggedCommandQueuing isenabledsincemore than 1 command can beoperated on concurrently. For instance,while a diskoperationis being performed for onecommandanothercommand can bereceived via theSCSI bus andplaced in thedevicecommandqueue. Certain environments maycauseCommandExecution Overhead toincrease if theadded function to process thequeue and themessages associated withqueuing is not permitted to overlapwith a diskoperation.
3.5.5.1 Reordered CommandsIf the Queue AlgorithmModifier Mode Parameterfield is set toallow it, commands in thedevicecommandqueue may beexecuted in adifferent order than theywere received. Commands arereordered sothat theSeek portion of Command Execution time is minimized. Theamount of reduction is a function of thelocation of the 1st requested block percommand and therate at which thecommands aresent to thedrive.
3.5.5.2 Back-To-Back Write Commands
If all of the requirements are met as stated in theULTRASTAR18XP/9LPSCSILogical Interface Specifica-tion section describingBack-To-Backwrite commands, contiguous data from 2 or moreconsecutive writecommands can bewritten to thedisk without requiring any diskLatency.
Note: There is aminimum write command transfer length for agiven environment where continuouswriting to the disk can not bemaintained withoutmissing a motor revolution. When Write Caching isenabled the likelihood isincreasedthat shorter transfer writecommands canfulfill the requirements neededto maintain continuouswriting to the disk.
Note: All times listed in this section are provided forinformation only sothat theperformance forotherenvironments/workloads can be approximated. Thesecomponenttimes should not be measuredagainst thespecification.
P8 End read disk transfer (Sector size)/(SCSIData TransferRate)P9 Transfer last few SCSI blocks in (5)(Sector size)/(SCSIData TransferRate)P10 SCSIread ending processing 2 µsP11 Status,Command CompleteMsg., BusFree 3 µs
Note: The Commandexecution overhead for a readcommand iscomposed of P1,P2(a&b),P3,P5,P10,andP11. (0.28 mstotal).
Time to Read data = P 1 + P 2 + P 3 + P 4 + P 5 + P 6 + P 7 + P 8 + P 9 + P 1 0 + P 1 1
3.7 Write Command PerformanceNote: This case is forRandomSCSI Write commands, with ReadAheaddisabled.
Note: All times listed in this section are provided forinformation only sothat the performance forotherenvironments can be approximated. Thesecomponenttimes should not be measuredagainst thespecifica-tion.
S1 Selection,Identify Msg., CDB 15 µsS2a SDP Msg. 1 µsS2b DisconnectMsg., BusFree 1 µsS3 Arbitrate, Reselect,Identify Msg. 6 µsS4 start SCSI transfer out 4 µsS5 SCSI busdata transfer out (Transfer size)/(SCSIData TransferRate)S6 End SCSItransfer out 4 µsS7A SDP Msg. 1 µsS7B DisconnectMsg., BusFree 1 µs
Note: The SCSI bus overhead for awrite command iscomposed ofS1,S2(a&b),S3,S4,S6,S7,S8 and S9.(0.04 mstotal).
3.7.1.2 Command Execution Overhead
P1 Selection,Identify Msg., CDB 15 µsP2a SDP Msg. 1 µsP2b DisconnectMsg., BusFree 1 µsP3 Start seek 258 µsP4 Seek (forexample, average seek) (Write Seek = 7.5 or 8.5 ms)P5 Set up write disk transfer 0 µsP6 Latency (for example,half revolution) (Latency = 4.17 ms)P7 Disk data transfer (Data transferred)/(TypicalData Sector TransferRate)P8 End write disk transfer 75 µsP9 SCSI write ending processing 25 µsP10 Arbitrate, Reselect,Identify Msg. 6 µsP11 Status,Command CompleteMsg., BusFree 3 µs
Note: The Commandexecution overhead for awrite command iscomposed of P1, P2(a&b), P3, P5, P8,P9, P10 and P11.(0.38 mstotal).
Time to Write data = P 1 + P 2 + P 3 + P 4 + P 5 + P 6 + P 7 + P 8 + P 9 + P 1 0 + P 1 1
3.8 Skew
3.8.1 Cylinder to Cylinder SkewCylinder skew is the sum of the sectorsrequired forphysically moving the heads, which is a function of theformatted block length andrecording density(notch #). Cylinder skew is always afixed minimum amount oftime and therefore thenumber ofsectorsvaries depending on whichnotch isbeing accessed and theblocklength. Theminimum amount of timerequired for acylinder switch is 2.13 ms.
3.8.2 Track to Track SkewTrack skew is the timerequired to perform aswitch between heads on the samecylinder. That time is .83ms.
3.9 Idle Time Function Considerations
The execution of various functions by thedrive during idle times may result indelays of commandsrequested bySCSI initiators. 'Idle time' is defined as timespent by thedrive not executing acommandrequested by aSCSI initiator. The functions performed duringidle time are:
1. PredictiveFailure Analysis (PFA)2. SaveLogs and Pointers3. Disk Sweep
The commandexecution time forSCSI commandsreceived whileperforming idle time activities may beincreased by theamount of time ittakes tocomplete theidle time activity. Arbitration, Selection, Messageand Commandphases, and disconnects controlled by thedrive are not affected by idle time activities.
Note: CommandTimeout Limits do not change due toidle time functions.
Following, are descriptions of the various types ofidle functions, how often theyexecute and theirduration.Duration isdefined to be themaximum amount of time theactivity can add to acommand when noerrorsoccur. No more than oneidle function will be interleaved with eachSCSI command. Following thedescriptions is asummary of thepossibleimpacts to performance.
There are mechanisms tolessenperformance impacts, and in somecases virtuallyeliminate those impactsfrom an Initiator'spoint of view.
1. Normal recommended operation
Idle Time Functions are onlystarted if thedrive has not received aSCSI command for atleast 5seconds. Thismeans thatmultiple SCSIcommands areaccepted and executedwithout delay if thecom-mands arereceived by the drivewithin 5 secondsafter the completion of aprevious SCSI command.This mechanism has thebenefit of not requiringspecialsystem software(such asissuing SCSI RezeroUnit commands at known &fixed time intervals) inorder to control if and whenthis functionexecutes.
2. Synchronizedoperation
Applications whichcannot accommodateinterruptions at all may consider synchronizingidle activitiesto the system needsthrough use of the TCC bit inMode Page 0h and theRezero Unitcommand.
Note: An example of this limitingmechanism's use would be if asystem isknown to issueSCSI com-mands for anapplicationgreater than 5 secondsapart and anIdle Time Functiondelay could not betolerated by the system on any ofthose commands. This wouldeliminatedrive initiated IdleTime Func-tion from even starting while thesystem/application is in acritical response time period of operation.
3. No PFA operation
Idle Time initiated PFA can be disabled by setting the"Perf" bit in Mode Page 1Ch. See theULTRASTAR18XP/9LPSCSILogical Interface Specificationfor details.
3.9.1 Predictive Failure Analysis (PFA)
PFA monitors drive parameters and can predict if adrive failure is imminent. There are "symptomdriven"PFA processeswhich occur duringError Recovery Procedures. The impacts of thoseupon perceived per-formance are not included heresince they areincluded in thenormal error recovery times,which are takeninto account by the"Typical Data Sector TransferRate".
There arealso "measurementdriven" PFA processeswhich occur duringIdle Time. Seven different PFAmeasurements are taken foreach head. The measurements for all heads aretaken over a period of 4hours,therefore the frequency of PFA isdependent on thenumber of heads a particularmodel has. Thedriveattempts tospread the measurements outevenly in time andeachmeasurementtakesabout 80milliseconds.
For example, a model with 20 heads will perform one PFA measurementevery 1.7 minutes (240 /(7*20)). Models with 10 heads perform ameasurement every 3.4 minutes.
For the lastheadtested for a particularmeasurement type (onceevery 34minutes), the data isanalyzed andstored. The extra execution time forthoseoccurrences is approximately 40milliseconds.
This measurement/analysisfeature can be disabled for criticalresponse time periods ofoperation bysettingthe Page 1ChMode Parameter PERF = 1. Theusing system also has theoption of forcing execution atknown times by issuing theSCSI Rezero Unitcommand if thePage 0hMode Parameter TCC = 1. Alltests for all headsoccur at thosetimes. See theULTRASTAR18XP/9LPSCSILogical Interface Specifica-tion for more detailsaboutPFA, PERF and TCC.
The drive periodically saves data in logs in the reservedarea of the disks. Theinformation is used by thedrive to supportvariousSCSI commands and for the purpose offailure analysis.
Logs areattempted to besaved every 26-35minutesduring idle time. Theamount of time ittakes toupdatethe logs variesdepending on thenumber oferrors since the lastupdate. Inmost cases,updating thoselogsand the pointers to thoselogs will occur in less than 30 ms.
3.9.3 Disk Sweep
The heads aremoved to anotherarea of the disk if thedrive has not received aSCSI command for atleast40 seconds. Afterflying in the samespot for 9 minutes withouthaving receivedanother SCSIcommand, theheads aremoved to another position. If no other SCSIcommand isreceived, theheads aremoved every 9minutesthereafter. Assoon as aSCSI command isreceived, theperiod for the 1st occurance is reduced backdown to 40 seconds. The period isincreasedback to 9 minutes for subsequentoccurances should nomoreSCSI commands bereceivedduring that time. Execution time isless than 1 fullstrokeseek.
3.9.4 Summary
Table 9. Summary ofIdle Time Function Performance Impacts
3.10 Command Timeout LimitsThe 'CommandTimeout Limit' is defined as the timeperiod from the SCSI Arbitration phasethrough theSCSI CommandCompletemessage, associated with aparticularcommand.
The following times are forenvironmentswhereAutomatic Reallocation isdisabled andthere are noqueuedcommands.
3.10.1.1 Reassignment Time
The drive should beallowed aminimum of 45 s tocomplete a"Reassign Blocks"command.
Idle Time Func-tion Type
Period ofOccurance(minutes)
Duration (ms) Mechanism toDelay
Mechanism toDisable
PFA 34/(trk/cyl) 80 Re-instructionPeriod, TCC
PERF
SaveLogs &Pointers
26 30 Re-instructionPeriod, TCC
Disk Sweep 2/3 - since lastcommand
17 Re-instructionPeriod
9 - since lastoccurance
Note: "Re-instruction Period" is the time betweenconsecutiveSCSI commandrequests.
18.2GB models should beallowed 110 minutes to complete a"Format Unit" command. If theVendorUnique ModePage 00h bitnamed"FFMT" is set equal to '1'b, then they should beallowed 60 seconds tocomplete.
9.1GB models should beallowed 55minutes to complete a"Format Unit" command. If theVendor UniqueMode Page 00h bitnamed"FFMT" is setequal to'1'b, then they should beallowed 30 seconds tocomplete.
3.10.1.3 Start/Stop Unit Time
The drive should beallowed aminimum of 45 s tocomplete a "Start/Stop Unit"command(with Immed bit= 0).
Initiators shouldalso use this time to allowstart-up sequences initiated by auto start ups and"Start/StopUnit" commands(with Immed bit = 1) tocomplete andplace the drive in a "ready for use"state.
Note: A timeout of one minute or more is recommended but NOTrequired. Thelarger systemtimeoutlimit allows the system to takeadvantage of theextensiveERP/DRPthat thedrive mayattempt inorder tosuccessfullycomplete the start-upsequence.
3.10.1.4 Medium Access Command Time
The timeout limit for medium accesscommands thattransfer user dataand/or non-user data should be aminimum of 30 s.Thesecommandsare:
Log SenseMode Select (6)Mode Sense (6)Pre-FetchRead (6)Read (10)ReadCapacityReadDefectData
Note: The 30 s limit assumes the absence of buscontention and user datatransfers of 64 blocks orless.This time should be adjusted for anticipated buscontention and iflonger data transfers are requested.
When AutomaticReallocation is enabled add 45 s to thetimeout of thefollowing commands;Read (6),Read(10), Write (6), Write (10), Write andVerify, and Write Same.
3.10.1.5 Timeout limits for other commands
The drive should beallowed aminimum of 5 s tocompletethesecommands:
InquiryRequestSenseReadBuffer
Start/Stop Unit(with Immed bit = 1)Synchronize CacheTest Unit Ready
The command timeout for a command that is notlocated at the head of thecommandqueue should beincreased by the sum ofcommandtimeouts for all of the commands that areperformed before it is.
A minimum of 2 mm clearance should begiven to thebottom surface except for a 10 mmmaximumdiam-eter areaaround thebottom mounting holes. A minimum of 1 mmclearance should begiven to thecoverflangesaround the topedge of thefile.
There should be 7 mm ofclearancebetweenULTRASTAR 18XP/9LP drives that are mounted withtheir top sides facingeachother.Drives from other manufactures mayrequire additionalspacing due to straymagneticfields.
Note: For propercooling it is suggestedthat a minimumclearance of 7 mm be providedunder thedriveand on top of the drive. Forfurther information see7.1.1, “Temperature Measurement Points” onpage 77.
4.3 Mounting Guidelines
The drive can bemounted with anysurface facingdown.
The drive is available with bothside andbottom mounting holes.Refer to Figure 17 onpage 46,Figure 18on page 47,Figure 19 onpage 48, andFigure 20 onpage 49 for thelocation of thesemounting holes foreach configuration.
The maximumallowablepenetration of the mounting screws is 3.8 mm.Screwslonger than 3.8 mm maycausepermanentdamage to thedrive.
The recommended torque to beapplied to themounting screws is 0.8Newton-meters + / - 0.2 Newton-meters. IBM will provide technicalsupport to usersthat wish to investigate highermounting torques intheir application.
For more information on mountingguidelines seesection7.5, “Drive Mounting Guidelines” on page 83.
4.3.1 SCA Mounting Guidelines
Since the SCAmounting systemlacks the compliant cabling of alternate connectors(50/68), the systemdesignermust nowconsider the followingmounting situations anddesign thesystem appropriately for longterm reliability. This list of guidelines is not intended to be exhaustive.
1. The SCA-2 connector should not berequired tosupport theweight of the drive.
2. Operational vibrationoccurring between the matinghalves of the SCAconnector should beavoided.
3. The drive should befirmly securedonce the connector mate hasoccured.
4. The connector wasdesigned to allow for'mismate' or offsetduring theplugging operation. Excessiveoffsets between thedrive connector and the backplanewill induce stress on theconnectorsystem andcard.
WARNING: The drive may be sensitive to usermounting implementation due toframe distortion effects.IBM will provide technicalsupport toassist users toovercomemountingsensitivity.
The DC power connectorsused on all models(50, 68 and 80 pinSCA-2) are an integralportion of the50/68pin SCSI 'Unitized' Connectors and the 80 pin'SingleConnector Attachment'(SCA-2) Connector.
50 pin models use an AMP connector (PN84412-1)that is compatible with theANSI SCSI "A" connectorspecifications.
68 pin models use a Molex connector (PN87360-0001)that is compatible with theANSI SCSI "P" con-nectorspecifications.
The 80 pin SCA-2 models use an AMP connector (PN5-917593-9)that iscompatible with theSpecification| of: 'Single Connector Attachment forSmall SCSI Disk Drives' SFF-8046document,revision 2.7. Place-| ment of theconnector is in compliance with the SFF-8337,revision 1.2.
The connector pinshave a plating of 30 micro-inches ofgold.
Power pinassignments for the 50 and 68 pin models areshown in Table 11.
Table 11. Power connector pinassignments
| Refer to the sections entitled2.4, “ Power Requirements” onpage 15,2.4.8,“Additional 5V Current| Requirements for HighVoltage DifferentialSCSI” on page 24 , and 2.4.9,“Additional 5V Current Require-| ments for LowVoltage Differential(Ultra 2)” on page 25, for details on drivepower requirements.
5.2 SCSI Bus Connector
ULTRASTAR 18XP/9LP has differentmodel types that support 50 or 68 pinSingle-Ended or 68 pinDifferential configurations. Also supported is the 80 pin SCA-2 in aSingle-Ended configuration. This
sectiondescribes the signalassignments of theULTRASTAR 18XP/9LPSCSI connectors.
These connectors have afinish metallurgy 30 micro-inchGold plating.
50 pin models use an AMP PN 84412-1connector. The connector is compatible with theANSI SCSI "A"connectorspecifications. It is limited to 8 bit data transfersonly. Refer to Figure 23 onpage 51 for a rearview of the 50 pinmodel connector.
The SCSI connector contactassignments for the 50 pin single-endedmodel are shown in Table 12 onpage 54.
68 pin models use a Molex connector (PN87360-0001)that is compatible with theANSI SCSI "P" con-nector specifications. It can transfer data in both 8 bit(narrow) and 16 bit(wide) modes. Refer toFigure 22 onpage 51 for a rearview of a 68 pinmodel.
| The SCSI connector contactassignments for the 68 pin single-ended and LVDmodels are shown in| Table 13 onpage 55.
The SCSI connector contactassignments for the 68 pindifferential models are shown in Table 14 onpage 56.
| Table 13. 68 PinSingle-Ended SCSI and LVDConnector ContactAssignments
| Signal Name (LVD Signal Name)| Connector Contact| Number| Signal Name (LVD Signal Names are the| same)
| GROUND (+DB(12))| 1| 35| -DB(12)
| GROUND (+DB(13))| 2| 36| -DB(13)
| GROUND (+DB(14))| 3| 37| -DB(14)
| GROUND (+DB(15))| 4| 38| -DB(15)
| GROUND (+DB(P1))| 5| 39| -DB(P1)
| GROUND (+DB(0))| 6| 40| -DB(0)
| GROUND (+DB(1))| 7| 41| -DB(1)
| GROUND (+DB(2))| 8| 42| -DB(2)
| GROUND (+DB(3))| 9| 43| -DB(3)
| GROUND (+DB(4))| 10| 44| -DB(4)
| GROUND (+DB(5))| 11| 45| -DB(5)
| GROUND (+DB(6))| 12| 46| -DB(6)
| GROUND (+DB(7))| 13| 47| -DB(7)
| GROUND (+DB(P))| 14| 48| -DB(P)
| GROUND (GROUND)| 15| 49| GROUND
| GROUND (GROUND)| 16| 50| GROUND
| TERMPWR (TERMPWR)| 17| 51| TERMPWR
| TERMPWR (TERMPWR)| 18| 52| TERMPWR
| OPEN (OPEN)| 19| 53| OPEN
| GROUND (GROUND)| 20| 54| GROUND
| GROUND (+ATN)| 21| 55| -ATN
| GROUND (GROUND)| 22| 56| GROUND
| GROUND (+BSY)| 23| 57| -BSY
| GROUND (+ACK)| 24| 58| -ACK
| GROUND (+RST)| 25| 59| -RST
| GROUND (+MSG)| 26| 60| -MSG
| GROUND (+SEL)| 27| 61| -SEL
| GROUND (+C/D)| 28| 62| -C/D
| GROUND (+REQ)| 29| 63| -REQ
| GROUND (+I /O)| 30| 64| -I/O
| GROUND (+DB(8))| 31| 65| -DB(8)
| GROUND (+DB(9))| 32| 66| -DB(9)
| GROUND (+DB(10))| 33| 67| -DB(10)
| GROUND (+DB(11))| 34| 68| -DB(11)
| Note:
| 8 bit SE devices whichconnect to theP-cable should tie thefollowing signals inactivehigh: -DB(8), -DB(9), -DB(10),| -DB(11), -DB(12), -DB(13), -DB(14), -DB(15),-DB(P1) or select "Disable Wide Negotiations" on theFront Option| JumperBlock and 'float' the samesignals.
| For 8 bit LVD devices in LVD or SEmode thefollowing signalsmust be tied inactive ( + = inactive low, - = inactive| high). +/-DB(8), +/-DB(9), +/-DB(10), +/-DB(11), +/-DB(12), +/-DB(13), +/-DB(14), +/-DB(15),+/-DB(P1).| Floating these signals in not sufficient.
The 80 pin SCA-2 models use an AMP connector (PN5-917593-9)that iscompatible with theSpecification| of: 'Single Connector Attachment forSmall SCSI Disk Drives' SFF-8046document,revision 2.7. Place-| ment of theconnector is in compliance with the SFF-8337,revision 1.2.
The connector has afinish metallurgy of 30 micro-inchGold plating.
Data transfers in both 8 bit(narrow) and 16 bit(wide) modes are supported.Refer to Figure 21 onpage 50for a rear view of an 80 pinmodel.
Signal Name Connector ContactNumber
Signal Name
+DB(12) 1 35 -DB(12)
+DB(13) 2 36 -DB(13)
+DB(14) 3 37 -DB(14)
+DB(15) 4 38 -DB(15)
+DB(P1) 5 39 -DB(P1)
GROUND 6 40 GROUND
+DB(0) 7 41 -DB(0)
+DB(1) 8 42 -DB(1)
+DB(2) 9 43 -DB(2)
+DB(3) 10 44 -DB(3)
+DB(4) 11 45 -DB(4)
+DB(5) 12 46 -DB(5)
+DB(6) 13 47 -DB(6)
+DB(7) 14 48 -DB(7)
+DB(P) 15 49 -DB(P)
DIFFSENS 16 50 GROUND
TERMPWR 17 51 TERMPWR
TERMPWR 18 52 TERMPWR
OPEN 19 53 OPEN
+ A T N 20 54 -ATN
GROUND 21 55 GROUND
+ B S Y 22 56 -BSY
+ A C K 23 57 -ACK
+ R S T 24 58 -RST
+ M S G 25 59 -MSG
+ S E L 26 60 -SEL
+ C / D 27 61 -C/D
+ R E Q 28 62 -REQ
+ I / O 29 63 -I/O
GROUND 30 64 GROUND
+DB(8) 31 65 -DB(8)
+DB(9) 32 66 -DB(9)
+DB(10) 33 67 -DB(10)
+DB(11) 34 68 -DB(11)
Note: 8 bit devices whichconnect to theP-cable should tie thefollowing signals inactive:+/-DB(8), +/-DB(9),+/-DB(10), +/-DB(11), +/-DB(12), +/-DB(13), +/-DB(14), +/-DB(15),+/-DB(P1). All other signals shall becon-nected as defined.
| Table 15. 80 PinSCA-2 SE and LVDConnector ContactAssignments
| Signal Name (LVD Signal Names are the| same)| Connector Contact| Number| Signal Name (LVD Signal Name)
| 12 V Charge| 1| 41| 12V Ground (12 VCharge)
| 12 Volt| 2| 42| 12V Ground (12Volt)
| 12 Volt| 3| 43| 12V Ground (12Volt)
| 12 Volt| 4| 44| Mated 1 (Mated 1)
| Reserved /NC| 5| 45| Reserved /NC (Diffsens)
| Reserved /NC| 6| 46| Ground (Reserved/NC)
| -DB(11)| 7| 47| Ground (+DB(11))
| -DB(10)| 8| 48| Ground (+DB(10))
| -DB(9)| 9| 49| Ground (+DB(9))
| -DB(8)| 10| 50| Ground (+DB(8))
| -I/O| 11| 51| Ground (+I /O)
| -REQ| 12| 52| Ground (+REQ)
| -C/D| 13| 53| Ground (+C/D)
| -SEL| 14| 54| Ground (+SEL)
| -MSG| 15| 55| Ground (+MSG)
| -RST| 16| 56| Ground (+RST)
| -ACK| 17| 57| Ground (+ACK)
| -BSY| 18| 58| Ground (+BSY)
| -ATN| 19| 59| Ground (+ATN)
| -DB(P0)| 20| 60| Ground (+DB(P0))
| -DB(7)| 21| 61| Ground (+DB(7))
| -DB(6)| 22| 62| Ground (+DB(6))
| -DB(5)| 23| 63| Ground (+DB(5))
| -DB(4)| 24| 64| Ground (+DB(4))
| -DB(3)| 25| 65| Ground (+DB(3))
| -DB(2)| 26| 66| Ground (+DB(2))
| -DB(1)| 27| 67| Ground (+DB(1))
| -DB(0)| 28| 68| Ground (+DB(0))
| -DB(P1)| 29| 69| Ground (+DB(P1))
| -DB(15)| 30| 70| Ground (+DB(15))
| -DB(14)| 31| 71| Ground (+DB(14))
| -DB(13)| 32| 72| Ground (+DB(13))
| -DB(12)| 33| 73| Ground (+DB(12))
| 5 Volt| 34| 74| Mated 2 (Mated 2)
| 5 Volt| 35| 75| 5 V Ground (5 V Ground)
| 5 V Charge| 36| 76| 5 V Ground (5 V Ground)
| Reserved| 37| 77| Active LED Out (Active LED Out)
| AUTO START| 38| 78| AUTO START DELAY (AUTO START| DELAY)
| -SCSI ID(0)| 39| 79| -SCSI ID(1) (-SCSI ID(1))
| -SCSI ID(2)| 40| 80| -SCSI ID(3) (-SCSI ID(3))
| Note:
| 8 bit SE devices whichconnect to the SCA connector should tie thefollowing signals inactivehigh: -DB(8), -DB(9),| -DB(10), -DB(11), -DB(12), -DB(13), -DB(14), -DB(15),-DB(P1) or select "Disable Wide Negotiations" on theFront| Option JumperBlock and 'float' the samesignals.
| For 8 bit LVD devices in LVD or SEmode, thefollowing signalsmust be tied inactive ( + = inactive low, - = inactive| high). +/-DB(8), +/-DB(9), +/-DB(10), +/-DB(11), +/-DB(12), +/-DB(13), +/-DB(14), +/-DB(15),+/-DB(P1).| Floating these signals is not sufficient.
Single-endedmodels permitcable lengths of up to 6 meters (19.68feet). It should be notedhowever thatusers who plan to use "Fast" data transfers withsingle-ended models shouldfollow all of the ANSI SCSIguidelines for single-ended"Fast" operations. This mayinclude a cable length ofless than 6meters.
When operating in Fast-20 modecable lengths of up to 3 meters(9.84 feet) aresupported.
SCA-2 connector models are notdesigned for direct cableattachment due to the combination of power andSCSI bus signals. "Fast" data transfers with SCA models shouldfollow all of the ANSI SCSI guidelines forsingle-ended"Fast" operations.
| High Voltage Differentialmodels permitcable lengths of up to 25 meters (82.02feet). Cablesmust meet the| requirements fordifferential cables as setforth in the Information Technology SCSIParallel Interface 2| (SPI-2) standard under"CableRequirements".
| Low Voltage Differentialmodels permitcable lengths of up to 12 meters(39.4 feet) whenoperating in LVD| mode. Cablesmust meet the requirements for LVDcables as setforth in the Information Technology SCSI| Parallel Interface 2(SPI-2) standard under"CableRequirements".
| The ANSI SCSI standardstatesthat any stubfrom main cablemust notexceed 0.1meters forsingle-ended
| or LVD cables and 0.2meters for high voltagedifferential cables.ULTRASTAR 18XP/9LP has a| maximuminternal stub length of 0.06 meters on allsingle-ended and LVDSCSI signals, and 0.10 meters on| all high voltage differentialSCSI signals. To remain compliant with the standard the SCSI buscable must| not add more than 0.05meters additionalstub length to any of the single-endedSCSI and LVDsignals or| 0.10 meters to any HVDSCSI signals.
5.2.5 SCSI Bus Terminators (Optional)
Upon request,SingleEnded 50 and 68 pin models areavailable with on cardSCSI busActive terminators.
For thosecards having the ActiveTermination feature, thisfunction can be enabled byinstalling a jumperbetween pins 13 and 14 of theFront Option JumperBlock or connecting pins 9 and 11 of theAuxiliaryConnector on 68 SCSI pinmodels. (Refer to Figure 26 onpage 67, andFigure 29 onpage 70.) The usingsystem is responsible for making surethat all requiredsignals areterminated at both ends of thecable. See5.2.7, “Single-endedSCSI BusElectrical Characteristics” on page 61 forinput capacitancevalues when ter-minators are disabled and whenterminators are not populated on thecard.
80 pin SCA-2 Models do nothave internalSCSI bus terminators.
Some externalterminatorpossibilities forsingleended cabled systems arelisted below: .
Table 16. SingleEndedSCSI Terminators
High Voltage Differentialmodels do not have internalSCSI bus terminators. Someexternalterminator pos-sibilities are listedbelow: .
| Low Voltage Differentialmodels do not have internalSCSI bus terminators. Someexternalterminator pos-| sibilities for Low Voltage Differential systems are listedbelow: .
| Table 18. LVD Terminators
5.2.6 SCSI Bus Termination Power
Termination power isoptionally provided forsystemsthat desire to use it. Inorder to use the terminationpower, the userneeds to install ajumper between pins 1 and 2 of theTermPower Block. (Refer toFigure 25 onpage 66,Figure 26 onpage 67.) Thejumper should only beinstalled on one device,whichshould be thelast device on theSCSI bus(i.e. the drivethat is physically closest to aterminator). 68 pinmodels can source up to 2.0Amps of current at 5.0 Volts ( + - 5%) fortermination power. 50 pin modelscan source up to 1.5Amps of current at 5.0 Volts ( + - 5%) fortermination power.
5.2.6.1 SCSI Bus Termination Power Short Circuit Protection
The ANSI SCSI specification recommends fordevices that optionally supply TERMPWR, to includecurrent limiting protection foraccidentalshort circuits. It alsorecommends that themaximumcurrentavail-able for TERMPWR should be 2 Amps. UL has adifferent requirementthat they call the 8 Amp rule.This rule statesthat when apower source leaves anenclosure(like SCSI TERMPWR in theSCSI cable), itmust trip 8 Amps of current within 1 minute.
The ULTRASTAR 18XP/9LP drive limits current to 2.0Amps thru the use of aresettable fusemounted on theelectronics card.
Systems may also provideshort circuit protection fordrive suppliedTERMPWR by limiting the current ofthe 5 Volt power itsupplies to the drive.
5.2.7 Single-ended SCSI Bus Electrical Characteristics
The following DC operatingcharacteristicspertain to thesingle-endedSCSI bustransceivers. All of these| parameters meet theInformation Technology SCSIParallel Interface 2(SPI-2) requirements.
Ta = 0 to 70 deg. C
Table 19. Single-Ended Bus Electrical Characteristics
| 5.2.8 High Voltage Differential SCSI Bus Electrical Characteristics
| High Voltage Differentialmodels meet all electrical requirements asdefined in theInformation Technology| SCSIParallel Interface 2(SPI-2) requirements.
| 5.2.9 Low Voltage Differential SCSI Bus Electrical Characteristics
| The following DC operatingcharacteristicspertain to the LowVoltage DifferentialSCSI bustransceivers.
| 5.2.9.1 Stub Length
| Minimum and Maximumtrace lengths, between theSCSI connector and the SCSIcontroller die pad, on the
| ULTRASTAR 18XP/9LPSCSI productcard are defined in thetable below.
| Table 20. LVD SCSI Controller to SCSIConnector Stub Length
| Measurement Parameter| Minimum| Maximum| SpecLimit| (Maximum)| Unit
| Stub Length| 32| 58| 100| mm
| 5.2.9.2 Capacitive Loading
| The test system used for capacitance measurements, is calibratedsuchthat total capacitance of the system is| zero at themating connector without a SCSIdrive pluggedinto the test system. The testconditions used for| measurement purposes are thosedefined in theSCSI Parallel Interface-2 (SPI-2) specification. The test| data provided is based on small lab samples.
| * Note: This out-of-spec condition was found on the68-pin product cards, and only on one differential| signalpair (DBP).
| 5.2.9.3 Driver Slew Rates
| The SlewRates (tRise, tFall) for Data and REQ aremeasured on a synchronous datatransfer. The testcondi-| tions and load circuits used formeasurement purposes are thosedefined in theSCSI Parallel Interface-2| (SPI-2) specification. The test dataprovided is based on small lab samples.
| Measurement Parameter| Minimum| Measured| SpecLimit| (Minimum)| Unit
| 80-Pin Connector (No Active Termination)
| tRise (Data)| 4.3| 1.0| ns
| tFall (Data)| 4.3| 1.0| ns
| tRise (REQ)| 3.3| 1.0| ns
| tFall (REQ)| 3.3| 1.0| ns
| 5.2.9.4 Assertion/Negation Periods
| The Assertion/Negation Periods for the REQsignal aremeasured on a synchronous datatransfer. The test| conditions and load circuits used formeasurement purposes are thosedefined in the SCSI Parallel| Interface-2 (SPI-2) specification. The test dataprovided is based on small lab samples.
| Measurement Parameter| Minimum| Measured| SpecLimit| (Minimum)| Unit
| 80-Pin Connector (No Active Termination)
| Assertion Period (REQ)| 9.3| 8.0| ns
| Negation Period (REQ)| 13.1| 8.0| ns
| 5.2.9.5 Data Setup/Hold Time
| The Data Setup andHold Times with respect to the REQsignal aremeasured on a synchronous data| transfer. The testconditions and load circuits used formeasurement purposes are thosedefined in theSCSI| Parallel Interface-2 (SPI-2) specification. The test dataprovided is based on small lab samples.
| Table 27. LVD SCSI Controller Single-Ended Setup andHold Time Data
| Measurement Parameter| Minimum| Measured| SpecLimit| (Minimum)| Unit
| 80-Pin Connector (No Active Termination)
| Setup Time (Data to REQ)| 15.8| 11.5| ns
| Hold Time (REQ to Data)| 23.1| 16.5| ns
| LVD SCSI Differential Test Results
| Table 28. LVD SCSI ControllerDifferential Setup andHold Time Data
| Measurement Parameter| Minimum| Measured| SpecLimit| (Minimum)| Unit
| 80-Pin Connector (No Active Termination)
| Setup Time (Data to REQ)| 10.8| 9.25| ns
| Hold Time (REQ to Data)| 10.6| 9.25| ns
| 5.2.9.6 Receiver Hysteresis
| Single-Ended Receiver Hysteresis is ameasurement of thevoltage levels atwhich the receiverchanges state| as defined in theSCSI Parallel Interface-2 (SPI-2) specification. The test dataprovided is based on small| lab samples.
5.3 Option Block Connector (Jumper Blocks)ULTRASTAR 18XP/9LP models contain a jumperblock that can beused to enable certain featuresand select the SCSI ID of thedrive. This jumper block is refered to as the'Front' Option JumperBlockdue to its location on thedrive (opposite the SCSI connector). This jumperblock varies in pindefinitionbased oninterface type (50, 68, Differential,SCA-2).
The OptionBlock connector(2x16) used on 50, 68 and 80 pin models is an AMPconnector (PN84156-5)having a pin spacing of 2mm.
The IBM Part Number for the 2mmjumpers is45G9800 and theTermination Power Enable jumperPartNumber is21H0793.
The 45G9800 PN is:-
2mm spacing,w/tab 8.5mmlong, connector is 3.5 mm long
Figure 29. Auxiliary Connector on the 68 pin Connector
Note: Either theFront Option Block "OR" the Auxiliary block may be used but notboth.
The 68 pin models contain an 'Auxiliary' connectorthat replicatessome of the functions contained in theFront Option JumperBlock. The Auxiliary connectorsignal definition conforms to the SCSI document:SFF-8009 Rev 2.0definition with the following exceptions:
1. EXTERNAL FAULT (XTFALT-) is not supported on pin 2
2. AUTO SPIN START waschosen as the 'vendorunique' signal assignment (on pin 4.) (Thissignal isan input to thedrive. The SCSI spec (SCSI SFF-8009)specifiesthis pin as anoutput.) This signalshould beuseful for those applicationsthat want to"auto-start" thedrive based on location dependentSCSI ID.
This pin should be handled in one of thefollowing ways:
a. tied toground (autospin start enabled)
b. allowed to'float' (no connection)
c. driven with anopencollector driver(>1mA sink capability)
5.3.2 SCSI ID (Address) PinsInformation on how to select aparticular address for theSCSI device ID is given inTable 30.
Note: In the addressdeterminationtables,"off" means jumper is not inplace and "on"means jumper is inplace.
Table 30. Address Determination of 68and 80 Pin Models
Table 31. Address Determinationof 50 Pin Models
5.3.3 Auto Start (& Delay) Pins
The Auto Start andAuto Start Delay pins control when and how thedrive can spin up and comeready.When configured forAuto-Startup, themotor spins up afterpower isapplied without the need of aSCSIStart Unit command. For no Auto-Startup, aSCSI Start Unitcommand isrequired to make thedrive spinand be ready for mediaaccessoperations. When in Auto-Startup mode, thedrive will delay itsstart time bya period of time multiplied by its ownSCSI address.Table 32 onpage 72 andTable 33 onpage 72showwhether or notAuto-Startup mode isactive and the delayperiods, where applicable, for allcombinations ofthe pins.
Table 32. 68p and 50p Auto-StartupModes selectable byAuto-Start/Delay Pin Combinations.. The"on" conditionindicatesthat a jumper is inplace to theadjacent groundpin. The "off" condition indicatesthat no jumperis in place.
Table 33. 80p Auto-StartupModes selectable byAuto-Start/Delay Pin Combinations. The"on" condition indicatesthat a jumper is inplace to theadjacent groundpin. The "off" condition indicatesthat no jumper is inplace.
5.3.4 External Activity (LED) Pins
The LED pins can be used todrive an external LightEmitting Diode. Pleaserefer to the LED pinsectionof the ULTRASTAR18XP/9LPSCSILogical Interface Specificationfor a detailed functional description ofthis pin.
Up to 33 mA ( + / - 5%) of TTL level LED sink current capability is provided.Current limiting for theLED is as shown in thefollowing diagram.
Pins (68 and 50 pin interface models) Drive Behavior
AUTO STARTDELAY
AUTO START Auto-Startup Mode ? Delay (sec) Multiplier
off off NO na
off on YES 0
on off YES 10
on on YES 4
Pins (80 interface pin models) Drive Behavior
AUTO STARTDELAY
AUTO START Auto-Startup Mode ? Delay (sec) Multiplier
If the Write Protect pin is jumpered to ground thedrive will prohibit SCSI commands thatalter the cus-tomer dataareaportion of themedia from being performed. The state of this pin ismonitored on a percommand basis. See theULTRASTAR18XP/9LP SCSI Logical Interface Specificationfor functionaldetails.
5.3.6 Disable Sync. Negotiation PinIf a Disable Target InitiatedSynchronous Negotiation pin is grounded then an Initiator isrequired to start anegotiation handshake if Synchronousand/or 'Wide' (DoubleByte) SCSI transfers are desired. Pleasereferto the ULTRASTAR18XP/9LPSCSILogical Interface Specificationfor more details onthis feature.
5.3.7 Disable SCSI Parity PinGroundingthis pin will disableSCSI Paritychecking.
5.3.8 Disable Unit Attention PinGrounding this pin will disable the drivefrom building Unit Attention Senseinformation for commandsimmediately following a Power On Reset(POR) or SCSI Bus Reset. Any pending UnitAttention condi-tions will also be cleared at POR orSCSI Resettimes.
5.3.9 Disable Wide Negotiations
Jumpering pin 31 to pin 32(refer to Figure 26 on page 67, Figure 27 on page 68, orFigure 28 on| page 69),will cause the 68 pin SE, 80 pinSCA, the 68 pindifferential, or the 80 pin and 68 pin Low| Voltage Differentialmodels to operate in a "Narrow'(Single Byte) mode. Thedrive will not negotiate for
| Upon request, 68 pinSingle Ended, or 50 pin models areavailable with on cardSCSI busActive termina-| tors.
For thosecards having the ActiveTermination feature, thisfunction can be enabled byinstalling a jumperbetween pins 13 and 14 of theFront Option JumperBlock or connecting pins 9 and 10 of theAuxiliaryConnector on 68 SCSI pinmodels. (Refer to Figure 26 onpage 67, andFigure 29 onpage 70.)
| 5.3.11 Force Single Ended Mode (LVD Only)
| Jumpering pin 23 to pin 24(refer to Figure 26 onpage 67, orFigure 27 onpage 68),will cause the 68 pin| and 80 pin SCA Low Voltage Differentialmodels to operate in aSingle Ended modeonly. The drive will| not use the DIFFSENSline to determine SE or LVD modes.
5.4 Spindle Synchronization
SpindleSynchronization is not afeature of theULTRASTAR 18XP/9LPdrives.
All detected errors excluding interface andBATs #1 (Basic Assur-ance Test)errors
Error detection ≥ 99%
FRU isolation = 100%
To the device when the"Recommended Initiator ErrorRecoveryProcedures" in theULTRASTAR18XP/9LP SCSI Logical Inter-face Specificationare followed.
No isolation to sub-assemblies within thedevice are specified.
6.2 Data ReliabilityProbability of not recovering data 10 in 1015 bits read
Recoverableread errors (Mean of the Population)10 in 1013 bits read(Measured at nominal DC conditions androomenvironment withdefault errorrecovery -QPE* enabled.)
With QPE enabled and the default thresholds, error reporting onlyoccursafter step 18.4
6.3 SPQL (Shipped product quality level)
All units are functionally tested immediatelyprior to packaging andshipment fromIBM. When subse-quently installed and functionally tested in anapprovedsystem,somedrives may notpass. In general, thepercentage ofdrives that fail will dependupon adherence to shipping and handlingguidelines, functional testcriteria and systemdesigncompatibility. Contact yourtechnicalsupportrepresentative for further informa-tion andassistance.
6.4 Failure Rate
This product isdesigned for use inapplications where highreliability and availability are critical. In general,actualfailure rateswill depend onusageconditions and systemdesigncompatibility.
Parameters such as ambient temperature,cooling air flow rate, relative humidity, ambientpressure (altitude),applied voltage, shock, vibration, on/off cycles and duty cycle will affectfailure rates. Failure rateprojections may only be determined fromdrive system testing.Contact yourtechnicalsupportrepresentativefor further information andassistance.
4 * Please reference QPE(qualify post error)definition in theULTRASTAR18XP/9LPSCSI Logical Interface Spec-ification.
It is recommended that thedrive does not remain inoperative forlonger than 180 days,especially if the shelfenvironment is at high temperature and humidity.
6.6 Start Stop Cycles
The maximum number ofstart stop cyclessupported is1800.
The IBM Corporatespecifications and bulletins,such as C-S1-9700-000 in thecontaminantssection,thatare referenced inthis document areavailable for review.
7.1 EnvironmentalTemperature
Operating Ambient 5 to 50˚C (41 to 122˚F)Operating Disk Enclosure 5 to 65˚C (41 to 149˚F)Storage 1 to 65˚C (34 to 149˚F) SeeNoteShipping -40 to 65˚C (-40 to 149˚F)
Temperature Gradient
Operating 20˚C (36˚F) perhour maximumShipping and storage below condensation
Humidity
Operating 5% to 90% noncondensingStorage 5% to 90% noncondensingShipping 5% to 95% (Applies at the packaged level)
Wet Bulb Temperature
Operating 26.7˚C (80˚F)maximumShipping and Storage 29.4˚C (85˚F)maximum
Elevation
Operating and Storage -304.8 to 3048meters(-1000 to 10,000 feet)Shipping -304.8 to 12,192meters(-1000 to 40,000 feet)
Note: Guidelines for storage below 1˚C aregiven in IBM TechnicalReport TR 07.2112.
7.1.1 Temperature Measurement Points
The following is a list ofmeasurement points andtheir temperatures.Maximum temperaturesmust not beexceeded at theworst case drive andsystem operating conditions with thedrive reading and writing at themaximumsystemoperations per secondrate.
Note: Figure 32 onpage 79defineswheremeasurements should be made to determine the topdisk enclo-sure temperatureduring drive operation. Figure 31 onpage 78defines themodulesthat arelocated on thebottom side of thecard and themeasurement location on thebottom of thedisk enclosure.
There must be sufficient air flow through the drive so that the disk enclosure andmodule temperaturemaximumlimits defined inTable 34 onpage 78 are not exceeded.
Figure 32. Temperature Measurement Point(top view)
7.1.1.1 Module Temp. Measurement Notes
1. Center on the topsurface of themodule.
2. If copper tape isused toattach temperaturesensors, it should be nolarger than 6 mmsquare.
7.2 Vibration and Shock
The operating vibration and shocklimits in this specification areverified in two mount configurations:
1. By mounting with the6-32 bottom holes with thedrive on 2 mm high by 10 mmdiameterspacers asrequired by section4.2, “Clearances” on page 44
2. By mounting on any two opposingpairs of the6-32 sidemount holes.
Other mount configurations may result indifferent operating vibration and shock performance.
7.2.1 Output Vibration Limits
Spindle residualimbalance not toexceed 0.5gram-millimeters for either 1.0 inch or 1.6 inch models.
7.2.2 Operating Vibration
The vibration is applied in each of the three mutually perpendicularaxes, one axis at atime. Referring toFigure 1 onpage 9, the x-axis is defined as a linenormal to thefront/rear faces, the y-axis isdefined as aline normal to theleft side/rightside faces, and the z-axis isnormal to the x-yplane.
WARNING All drives are sensitive torotary vibration. Mounting within using systemsshould minimize the rotationalinput to the drive mounting points due toexternal vibration. IBMwill provide technicalsupport toassist users toover-come problems due to vibration.
RandomVibration
For excitation in the x-direction and the y-direction, thedrive will operate without harderrors when sub-jected tovibration levels notexceeding the V4vibration level definedbelow.
For excitation in the z-direction, thedrive will operate without harderrors when subjected to vibrationlevelsnot exceeding the V4Svibration level definedbelow.
Note: The RMS value in the table below isobtained bytaking the squareroot of the area defined by theg² /hz spectrum from 5 to 500 hz.
The drivewill operate without harderrors when subjected to the sweptsine vibration of 1.0 G peak from 5to 300 hz in the x, y, and z directions.
This measurement is takenduring a frequency sweepfrom 5 to 300 hz and back to 5 hz. Thesweep ratewill be one hz persecond.
Note: 1.0 G acceleration at 5 hz requires 0.78 inchdouble amplitudedisplacement.
7.2.3 Nonoperating Vibration
No physicaldamage or degradedthroughputwill occur as long as vibration at the unpackageddrive in allthree directionsdefined above does notexceed the levelsdefined in thetable below. This measurement isperformed bysweepingfrom 5 hz to 300 hz and back to 5 hz at asweep rate of eightdecades perhour.
7.2.4 Operating Shock
No permanentdamagewill occur to thedrive when subjected to a 10 Ghalf sine waveshock pulse of11 millisecondsduration.
No permanentdamagewill occur to thedrive when subjected to a 10 Ghalf sine waveshock pulse of2 millisecondduration.
The shockpulses are applied in each of three mutually perpendicularaxes, one axis at atime.
For both the 1.0" and the 1.6" models, no harderror will occur if the unpackageddrive is subjected to a20 millisecondssquare pulse shock of 35 Gs orlessapplied to all threeaxes, onedirection at a time.
For the 1.0" models, no harderror will occur if the unpackageddrive is subjected to a 2 millisecondshalfsine pulseshock of 150 Gs orlessapplied to all threeaxes, onedirection at a time.
For the 1.6" models, no harderror will occur if the unpackageddrive is subjected to a 2 millisecondshalfsine pulseshock of 140 Gs orlessapplied to all threeaxes, onedirection at a time.
RotationalShock
The actuatorwill remain latched in thedisk landingzone if the unpackageddrive is subjected to a 2millisec-onds half sinepulse shock of15,000 radians per second squared orlessapplied to all threeaxes, one direc-tion at atime.
7.3 Contaminants
The corrosive gasconcentrationexpected to be typicallyencountered isSubclass G1; the particulateenviron-ment isexpected to be P1 of C-S1-9700-000(1/89).
Additionally, the populationaverage of thesoundpressure measured onemeter above thecenter of thedrivein idle mode will not exceed 36dBA.
Additionally, thepopulationaverage of thesoundpressure measured onemeter above thecenter of thedrivein idle mode will not exceed 41dBA.
Notes:
1. The above octaveband andA-weighted sound powerlevels arestatistical upper limits of the soundpower levels. See C-B 1-1710-027 and C-S 1-1710-006 forfurther explanation.
2. The drives aretested after aminimum of 20minutes warm-up inidle mode.
3. The operating mode is simulated byseeking at arate of 32seeks persecond.
4. The values for asamplesize of 5 orgreaterwill be less than orequal to the statedupperlimits with 90%confidence.
For the 1.0" models, no degradation in A-weightedidle sound powerwill occur if the unpackageddrive islimited to a 2 milliseconds half sinepulse shock of 150 Gs orlessapplied in theaxial direction (z axis), or300 Gs or lessapplied in radial direction (x-y plane).
For the 1.6" models, no degradation in A-weightedidle sound powerwill occur if the unpackageddrive islimited to a 2 milliseconds half sinepulse shock of 70 Gs orless applied in theaxial direction (z axis), or150 Gs or lessapplied in the radial direction (x-y plane). TheaverageA-weighted idle sound powerwillincrease by 0.3 Bels if theunpackageddrive is subjected to a 2 millisecondshalf sinepulse shock of 110 Gsapplied in theaxial direction (zaxis), or 210 Gsapplied in the radial direction (x-y plane).
Upper Limit Sound Power Requirements(Bels) for 1.0 inch Models
OctaveBand Center Frequency (Hz) A-weighted
125 250 500 1K 2K 4K 8K 16K Bel
Idle 4.4 3.2 3.2 3.4 4.0 4.1 3.8 3.8 4.4
Operating 4.4 3.2 3.2 3.9 4.2 5.2 4.5 4.2 5.4
Upper Limit Sound Power Requirements(Bels) for 1.6 inch Models
7.5 Drive Mounting Guidelines1. Use of the extremeside mounts will align the driveCenter of Gravity(CG) closer to thecenter of
stiffness. This willminimize off axis coupling and in-plane yawrotation about thespindleaxis.
2. Orient thespindleaxis parallel to thedirection ofminimum shockloading.
3. The carrier should notallow the drive to rotate in theplane of thedisk.
If any isolation between thedevice and theframe is to be used,it can be soft in thex,y,z, pitch androllaxis but should bestiff for the yaw axis. Yaw motion is rotation about thespindle axis which couplesdirectly into offtrack.
If isolators are used, they should provide naturalfrequenciesabout 25%lower than themotor speed.The idea is to place therigid body modesbelow primary excitationfrequencies and drivestructuralmodes. Isolatorsmust be well damped and ofsufficient strength so theywill not be torn by high non-operating shocks.
Otherwise, keep therigid body resonances of thedrive awayfrom harmonics of thespindle speed.
7200 RPMharmonics: 120Hz, 240Hz, 360Hz, 480Hz....
4. It is desirablethat thecarrier be asstiff as possible while allowingroom for theisolator mounts (if used).Rather thencreating a weak carrierthat flexes to fit thedrive, hold themounting gap totighter toler-ances. Aflexible carrier maycontain resonancesthat causeoperational vibrationand/or shock prob-lems.
5. If isolators are to be used,design formaximumsway. Adequateclearancearound alledges arenecessaryfor cooling andshock impacts. Maximum sway is usually determined bygeometry and compressibilitylimits of the isolatorgrommetplus somecarrier/rackflexibility. Metal to metal impactsmust beavoidedbecause they result inshort duration, high impactsloads; such waveforms canexcite high frequencymodes of the componentsinside the drive.
6. To minimize acoustic radiation,mount drives sothere is no"line of sight" between adrive anduser.
7.6 Drive/System Compatibility
ULTRASTAR 18XP/9LP drives aresupplied to using systemsthat demonstrate alevel of drive/systemcompatibility to thisspecification.
Verification prior to a formal system qualification isrecommended to determine whether thedrive/system iscapable ofachieving thequality andreliability requirements found inthis specification.
Preliminary testing toverify compatibility may be performed usingcommon laboratory instrumentationequipped with the appropriate transducer (thermal, power, shock, vibration andacoustics). Final verificationmust beperformed by measuring functional performance (error rates) of thedrive when installedwithin thesystem.
The following sections describe theparameters to beverified prior to and as a part of thesystem qualifica-tion test in order toachieve thequality andreliability requirements set forth by thisspecification.
Power The systemmust becapable of providingadequate power to thedrive as described inSection2.4, “ Power Requirements” onpage 15 Inaddition to voltage,current andcapacitance, the systemmust be capable of remaining within regulation when themaximum number ofdrives are installed in thesystem.
Specialconsiderationmust begiven to hot plug and differentialdrive/systemdesigns.Refer to 2.4.7, “'Hot Plug/Unplug' support” onpage 23 of this specification forrequirements andguidelines.
Thermal The systemmust supply adequatecooling and air flow to maintain casting andmodule temperaturelisted in Table 34 onpage 78. Operating Limitsmust be inaccordance with theTemperature measurement points shown inFigure 32 onpage 79 andFigure 31 onpage 78. The systemmust demonstratesufficient coolingto operate below the recommended temperatures for anygiven location that thedrivemay be installedwithin the system.
Special consideration forminimum clearancesmust be given to achieveadequatecooling of the drive.
Shock (Operating and Non-Operating)The systemmust maintain anenvironment that iscompatible with operating andnon-operating shockspecificationsfound in sections 7.2.4,“Operating Shock” onpage 80 and 7.2.5,“Nonoperating Shock” onpage 81. Both operating andnon-operating shock should be measured in all 3planes andfound to be within thelimitsset in this specification.
Vibration (Operating and Non-Operating)The systemmust maintain an environment that iscompatible with the operatingvibration specification found in section 7.2.2, “Operating Vibration” on page 79.Randomvibration must bemeasured in all three planes andfound to be compatiblewith the vibrationlevel in Table 35 onpage 80. Swept SineVibration must also bemeasured and be compatible with7.2.2,“Operating Vibration” onpage 79.
To achieve systemcompatibility for vibration, it is recommendedthat the systemconform tosection 7.5, “Drive Mounting Guidelines” on page 83.
Also, drives are sensitive torotary vibration. Mounting within using systemsmustminimize the rotationalinput to drive mounting points due toexternal vibration.
Electromagnetic Compatibility(EMC)The systemmust bedesigned toinsure that stray fields are notplacedclose to thedevice.Minimum clearancesmust bemaintained. Clearanceguidelines arefound insection4.2, “Clearances” on page 44.
Electrostatic Discharge(ESD)The drive contains electrical componentssensitive to ESD. System design andassembly process,must protect thedrive and must beverified to conform the pro-tection, care and handling guidelinesfound in section 7.10, “ESD Protection” onpage 87.
Interface CompatibilityThe drive/system, inconjunction withassociatedoperatingsoftware,must becapableof conforming to the pin configurations,cabling, command andtiming parametersfound in section5.0, “ElectricalInterface” on page 53.
Verification of the precedingparameters is recommended prior to starting a system test or qualification.Most parameters may beverified by using common laboratory instrumentation or simple inspection ofdesign,handling and process. For furtherinformation regarding verification testing, pleasecontact yourtech-nical supportrepresentative.
Final verification of drive/systemcompatibility must be determinedthrough functional testing. Adequatesystem testingmust beperformed to demonstrate conformance to theData Reliability requirements,refer-ence 6.2,“Data Reliability” on page 75.
7.7 Recommendations for Handling of Disk DrivesDisk Drives arevery fragile and can bedamaged ifdropped or impactedagainstanotherobject. Amount ofdamage to thedrive will depend on magnitude and duration of the impact. People handling the disk driveshould be trained in theproper handlingprocedures. Manufacturingprocesses,equipment, and DiskDriveholding containers/fixtures should be characterized and qualified toless than 50 G's in themanufacturingenvironment. Thefollowing are things to consider in thehandling andprotection of thedisk drive.
Damage may be caused by:
Dropping adrive onto ahard surface, evenover small distances
Drives mayfall over after being set onedge
Tapping adrive with a screw driver tip orother hard implement
Tapping adrive into position wheninstalling into a userframe
7.8 Breather Filter HoleUnder nocircumstances should the Filter BreatherHole be obstructed orlabels placedover the hole.
Figure 33. Breather Hole forFilter
7.9 Periodic MaintenanceNone required
7.10 ESD Protection
The ULTRASTAR 18XP/9LP disk drives contain electrical componentssensitive todamage due toelectrostatic discharge (ESD). Proper ESDproceduresmust befollowed during handling, installation, andremoval. Thisincludes the use of ESD wriststraps and ESD protective shipping containers.
Precautions such asusing ESDmats, wrist straps and grounding all surfacesthat areallowed to touch orcome close to the device arerecommended.
Known ESD dangerssuch aswalking across acarpet carrying thedevice should be avoided. It isrecom-mended that thedevice is alwaysstored in its anti-static package until it is ready for installation.
The product is approved as aRecognizedComponent for use inInformation TechnologyEquipmentaccording to UL 1950 Standard, third edition(without any D3deviations). The UL Recognized Com-ponentmarking is located on theproduct.
UL E133560 Vol 19 Section 1
CANADIAN STANDARDS ASSOCIATION (CSA) APPROVAL:
The product iscertified to CAN/CSA-C22.2 No.950-M95 Third Edition (without any D3deviations).The CSA certificationmark is located on theproduct.
LR-34074C
FLAMMABILITY REQUIREMENTS
Printedcircuit boards and all foam and otherplastic materials are ULRecognizedV-1, HF-1, or VTM-1or better. Smallplasticpartsthat will not contribute to afire will meet V-2flame class.
SAFE HANDLING:
The product is conditioned for safehandling inregards tosharpedges andcorners.
ENVIRONMENT:
IBM will not knowingly or intentionally ship any units which duringnormal intended use orforeseeablemisuse,would expose the user to toxic, carcinogenic, or otherwisehazardous substances atlevels abovethe limitationsidentified in thecurrent publications of the organizationslisted below.
InternationalAgency forResearch on Cancer(IARC)
National Toxicology Program(NTP)
OccupationalSafety andHealth Administration(OSHA)
American Conference ofGovernmentalIndustrial Hygienists(ACGIH)
California Governor'sList of Chemical Restricted under California SafeDrinking Water and ToxicEnforcement Act1986(Also known as California Proposition 65)
SECONDARY CIRCUITPROTECTION REQUIRED INUSING SYSTEMS
Care has beenexercised to not use anyunprotected components orconstructionsthat areparticularlylikely to causefire. However, adequate secondary overcurrentprotection is theresponsibility of the userof the product. Additional protectionagainst the possibility of sustainedcombustion due tocircuit orcomponentfailure mayneed to be implemented by the user withcircuitry external to theproduct. Over-currentlimits of the voltageinto the file of 10 amps orless should besufficient protection.
Pertaining to theULTRASTAR 18XP/9LPdisk drive, IBM will provide technicalsupport toassistusers in complying with theUnited StatesFederal Communications Commission(FCC) Rules andRegu-lations, Part 15, Subpart B Digital Devices "Class A and B Limits" . Tests for conformance to thisrequirement are performed with thedisk drivemounted in theusing system.
CISPR 22Requirements
Pertaining to theULTRASTAR 18XP/9LPdisk drive, IBM will provide technicalsupport toassistusers in complying with the Comite International Special des Perturbations RadioElectriques(International Special Committee on Radio Interference)CISPR 22 "Class A and B Limits" .
EuropeanDeclaration of Conformity.
Pertaining to theULTRASTAR 18XP/9LPdisk drive, IBM will provide technicalsupport toassistusers in complying with theEuropean Council Directive 89/336/EEC so the finalproduct cantherebybear the "CE" Mark of Conformity.
This is obtained byintegrating the drives in an IBMproduct. Productsintegrating thesedrives in alternativeenclosureswill still need to test the system to ensure it complies with the EuropeanDirective.